hello barton,
> Yikes! This seems to say that atanh is continuous along the (-inf,-1]
> from the lower half plane. I hope that I'm wrong, but as I interpret
> the CLHS, it is contradictory.
so far as i can tell, CLHS is indeed inconsistent on this point.
too bad about that.
i've convinced myself the One True Reading is to take
the formulas for trig functions as the spec,
and the stuff about branch cuts as commentary.
(it certainly can't work the other way around.)
i've committed a modified src/trigi.lisp to the 5.9.3 branch
(should be visible at http://cvs.sf.net/viewcvs.py/maxima/maxima/src
in a few hours) which defines acos, asin, and atanh functions
according to the CLHS formulas and uses those in place of
cl:asin, cl:acos, and cl:atanh for SBCL.
with that modified version, the rtest_trig problems with
float functions go away. there is still a problem with bigfloat
atanh (since bigfloat code isn't affected by the change i made).
maybe someone can take a look at src/trigi.lisp and
make some comments about that.
hth
robert